Guide
How to measure first-response time the right way
Most businesses measure response time in a way that misleads them. Why to count business hours only, why the median beats the average, and what a daily summary should include.
“We reply fast” is something almost every business says about itself. Once you start measuring, the picture can look different. And even when you do measure, it’s easy to get it wrong: counting overnight hours, trusting an average that one conversation can skew, or treating an auto-reply as an answer. Here’s how to measure first-response time without fooling yourself, and what to do with the numbers.
What you’re actually measuring
First-response time is the time between a customer reaching out and getting a real reply. Simple enough, but it involves two decisions:
- When the clock starts. At the customer’s first message in a new conversation.
- What counts as a reply. A reply from a person on your team or from an AI agent that actually addresses the question. An automatic “we got your message” doesn’t count, because the customer still has no answer.
If you use an AI agent, look at two numbers: time to the first reply of any kind, and time to the first human reply in conversations that were handed to the team.
Count business hours only
Say you’re open 9 am to 6 pm. A lead comes in at 10 pm and gets a reply at 9:05 the next morning. On a plain clock, that customer waited over eleven hours. In reality, your team replied five minutes after starting work.
Measuring on a plain clock mixes up two different questions:
- How fast does the team reply while it’s working?
- What happens to messages that arrive while you’re closed?
Both matter, but you need to see them separately. Measure response time within business hours only, and track separately how many leads came in after hours and what happened to them. If a lot arrive in the evening, that’s a sign those hours need coverage, for example by an AI agent.
The median beats the average
An average sounds sensible, but one forgotten conversation can throw it off completely.
An example: you answered nine leads in 3 minutes each and one after 5 hours. The average comes out at over half an hour, which describes none of those conversations. The median, the middle value when you sort all the times from shortest to longest, comes out at 3 minutes. It reflects what most customers actually experienced.
So what about the lead that waited 5 hours? Don’t hide it. Look at it on its own, as a delayed conversation, and find out why it happened.
If a few of these show up every week, look for a pattern: the same channel, the same time of day or the same person. Usually the cause is simple, like a channel nobody owns.
Unanswered conversations matter more than any average
Response time is only calculated on conversations that got a reply. A conversation nobody answered never enters the calculation, so it never hurts the number. That’s the danger: you can show a great response time while losing customers.
So also track:
- How many conversations are waiting for a reply right now, and how long each has waited.
- How many were never answered by the end of the day.
- Who owns them. A conversation with no assigned owner is the one most likely to be forgotten.
What else is worth measuring
Response time is a start, but it doesn’t tell you whether the conversation led anywhere. A few metrics that fill in the picture:
- How many conversations became sales.
- How many payment links were sent and not paid.
- Which channel and which campaign the closed conversations came from.
- How many conversations the AI agent resolved on its own, and what each one cost.
- Response time per team member, to see who’s overloaded and who needs help.
What a daily summary should include
A manager shouldn’t need to check a dashboard every hour. A short daily summary near the end of the working day is enough to know what happened and what needs attention. A good one covers:
- How many new conversations came in, and from which channels.
- Median first-response time within business hours.
- A list of conversations still unanswered.
- How many conversations turned into sales today.
- Open payment links worth following up.
- Conversations the agent handed to the team that nobody has picked up yet.
The rule: a summary should lead to action. If you finish reading it and don’t know who to call tomorrow morning, it isn’t doing its job.
If you only have a minute
If you take one thing from this: an unanswered conversation never shows up in your average, so check those separately. Beyond that, business hours only, and the median. In Chatles, the dashboard measures first-response time only within your business hours, and every day at 6 pm the manager gets an email with the day’s summary and the conversations still waiting.
